Tumor Necrosis Factor-α Receptor 1 Is a Major Predictor of Mortality and New-Onset Heart Failure in Patients With Acute Myocardial Infarction
The Cytokine-Activation and Long-Term Prognosis in Myocardial Infarction (C-ALPHA) Study
Background— Tumor necrosis factor alpha-α (TNF-α) activation is an independent prognostic indicator of mortality in patients with heart failure (HF). Despite the recognition that several TNF family cytokines are elevated during myocardial infarction, their role in predicting subsequent prognosis in these setting remains poorly understood.
Methods and Results— We performed a systematic evaluation of TNF-α and its type 1 and 2 soluble receptors, together with interleukin (IL)-6, IL-1 receptor antagonist, and IL-10, in 184 patients (132 men; mean age, 64±12) consecutively admitted for myocardial infarction. We correlated their values to short- and long-term incidence of death and HF (primary outcome). In 10 patients, we also studied the presence of transcardiac gradients for TNF-α and its soluble receptors. The control group comprised 45 healthy subjects who were sex and age matched (33 men; mean age, 65±6 years) to the patients. All tested cytokines were increased in patients, and no transcardiac or systemic AV difference was found. After a median follow-up of 406 days (range, 346 to 696 days), 24 patients died and 32 developed HF. Univariate analysis showed that all cytokines were related to outcome, whereas after adjustment for baseline and clinical characteristics, sTNFR-1 remained the only independent predictor of death and HF (hazard ratio, 2.9; 95% CI, 1.9 to 3.8, tertile 1 versus 3), together with left ventricular ejection fraction, Killip class, and creatine kinase-MB at peak.
Conclusions— sTNFR-1 is a major short- and long-term predictor of mortality and HF in patients with acute myocardial infarction.
Received June 16, 2004; revision received October 18, 2004; accepted November 1, 2004.
Despite best-available treatment, the cumulative incidence of mortality and heart failure (HF) increases time dependently in patients with previous myocardial infarction (MI).1 Furthermore, the predictive value of the clinical profile at the time of the necrotic injury remains limited.2 Consequently, to better identify patients at high risk for cardiovascular events after MI, new predictive markers should be established.
Several lines of evidences support a role for tumor necrosis factor alpha-α (TNF-α) and its soluble receptors as mortality predictors in patients with HF,3–5 and more recent data suggest that proinflammatory cytokine elevation antedates the development of HF in patients without prior MI.6 Several cytokines, particularly those belonging to the TNF family members, are known to be elevated during MI,7 but their role in predicting subsequent prognosis remains to be addressed. Therefore, our primary aim was to undertake a systematic evaluation of TNF-α and its type 1 (sTNFR-1) and 2 (sTNFR-2) soluble receptors, together with interleukin (IL)-6, IL-1 receptor antagonist (ra), and IL-10, in a consecutive series of MI patients and to correlate them to short- and long-term incidence of death and HF. In a subset of patients, we also studied the presence of transcardiac gradients for TNF-α and its soluble receptors.
Cytokines Activation and Long-Term Prognosis in Myocardial Infarction (C-ALPHA) study is an observational, ongoing, prospective, single-center investigation aimed at evaluating the predictive value of inflammatory cytokines with respect to mortality and HF in patients with acute MI.
The study population, enrolled between January 2001 and June 2003, consisted of 184 patients (of 366 screened) and 45 healthy subjects. Their clinical and biochemical profiles are presented in Table 1. The procedures followed were in accordance with institutional guidelines, and local ethics committee approval was obtained. All participants gave written informed consent.
The patients (n=184; 132 men) were 32 to 92 years of age (mean age, 64±12 years) and were consecutively admitted to hospital for prolonged (>20 minutes) chest pain accompanied by ST-segment changes of ≥1 mm in ≥1 peripheral leads of the ECG or of ≥2 mm in ≥1 precordial leads. Exclusion criteria were as follows: symptom onset >14 hours before hospital admission, Killip class 4, history of HF, presence of any known ongoing infectious disease, and any known current or past neoplastic or immunological disorder.
Occurrence of MI was confirmed in all patients by a 2-fold upper limit of normal (190 U/L) rise in creatine kinase (CK), with an increased level of CK-MB.
Of the 151 patients (82% of total) who presented with ST-segment elevation MI (STEMI), 114 (75%) were submitted to a primary percutaneous coronary intervention (PCI) at entry. Glycoprotein IIb/IIIa inhibitors were administered in 93 of these patients (82%) as adjunctive therapy during primary PCI.
Medical treatment at discharge in the patient population consisted of nitrates (97%), ACE inhibitors (90%), β-blockers (72%), statins (80%), aspirin (89%), thienopyridines (74%), diuretics (34%), angiotensin II receptor blockers (18%), and warfarin (5%). Dosages and timing of drug administration were in accordance with current AHA/ACC guidelines.
Echocardiographic investigation was performed within 3 days of hospital admission and at discharge, and left ventricular ejection fraction (LVEF) is given as the mean between the 2 measurements. The percentage of the extent of wall-motion abnormalities was obtained by dividing the number of akinetic, dyskinetic, and aneurysmal segments by the total number of segments evaluated (n=16). In 10 STEMI patients (age, 62±8 years; 8 men) who were not submitted to primary PCI for late (>12 hour) presentation, left and right cardiac catheterization was performed within a mean±SD of 2±1 days, and the systemic and transcardiac AV gradients for TNF-α, sTNFR-1, and sTNFR-2 were determined.
The control group comprised 45 healthy blood donors who were sex and age matched (33 men; mean age, 65±6 years) to the patients. None had clinical signs of acute or chronic illness or was receiving any treatment.
Blood Sampling and Processing
Antecubital venous blood was collected from all patients at entry (14±9 hours after symptom onset; range, 0.5 to 34 hours) and from all healthy subjects, left in ice for 45 minutes, and then centrifuged at 1700g at 4°C for 15 minutes. Serum obtained was stored at −80°C.
TNF-α, sTNFR-1, sTNFR-2, IL-6, and IL-1ra were measured as previously described,3,7 with sensitivity and intra-assay and interassay variabilities of 3 pg/mL, <5.2%, and <9.9%; 3 pg/mL, <2.9%, and <8.8%; 1 pg/mL, <2.5%, and <5.1%; 0.7 pg/mL, <4.2%, and <6.4%; and 22 pg/mL, <3.1%, and <5%, respectively. IL-10 was quantified by Quantikine Immunoassay by R&D with an intra- and inter-assay variabilities of <5% and 8%, respectively, and a sensitivity of <3.9 pg/mL.
C-reactive protein (CRP) was measured in patients (43±18 hours after symptom onset) and subjects by nephelometry from fresh serum according to the method of Behring Diagnostic.
Patients underwent outpatient visits every 6 months. A minimum follow-up of 12 months was planned; in 4 patients (2%), this was achieved by telephone contact. Two patients (1%) were lost to follow-up and were not included in the out-hospital analysis.
The primary outcome of the study was a hierarchical composite of total mortality and new-onset HF. Their incidence in our study population was estimated in accordance with the Gruppo Italiano per la Sopravvivenza nell’Infarto Miocardico (GISSI)-3 database because of the almost-identical inclusion criteria. For explanatory analysis, the predictive value of tested inflammatory cytokines has also been tested in relation to mortality alone. According to recent experience by our group,8 it was hypothesized that patients with high (above median value) and low (below median values) levels of TNF-α would display a cumulative incidence of the primary outcome of 66% and 84% at 1 year, respectively. Therefore, through the use of Freedman’s calculation, a final population of ≥165 patients was needed to detect a 40% hazard rate reduction in the composite primary outcome, with a 1-sided α error of 0.05, a power of 0.85, and a prespecified 10% of patients lost to follow-up.
All deaths were considered to be of cardiac origin unless a noncardiac origin was established clinically or at autopsy. Occurrence of HF, in accordance with previously proposed criteria,9 required the presence of rest or effort dyspnea and ≥1 of the following: pulmonary rales at lung auscultation, S3 tone, evidence of pulmonary congestion at chest x-ray, new appearance of peripheral edema, or use of diuretics.
Data are shown as mean±SD. Comparisons between patients with and without primary outcome were performed with Student’s t test or the Mann-Whitney test for nonparametric variables. Fisher’s exact test was used for categorical variables. Correlations between variables were tested by Pearson’s analysis. Survival curves were generated by the Kaplan-Meier method, and survival among groups was compared by use of the log-rank test. The prognostic value of increased cytokine and cytokine receptor levels was examined with a Cox proportional-hazards model. All variables included in Tables 1 through 3⇓⇓ were tested as univariate predictors of primary outcome. Given the limited number of events in relation to the number of studied parameters, a variable selection using the Akaike information criterion was performed, followed by a bootstrapped variance estimation to avoid overfitting.10 To compare the predictive role of different values of the studied parameters, receiver-operating characteristics (ROC) with their area under the curve (AUC) were constructed. The best prognostic cutoff for freedom from death and HF was defined as the one that maximized both sensitivity and specificity. To contrast prognostic accuracy, statistical comparison of AUCs was performed. Probability was significant at a level of P<0.05. Statistical analysis was performed with Statistica 6.1 (Statsoft) and R-language (R Foundation).
Characteristics of the studied population are shown in Table 1. Twenty-one patients (11%) and 5 patients (3%) were in Killip class 2 and 3 at entry, whereas 17 (7%) and 32 (17%) had a previous coronary revascularization procedure and history of MI, respectively. The site of MI was anterior in 79 patients (43%).
Cytokines and Cytokine Receptors
Cytokine and cytokine receptor distribution across the study population is presented in Table 2. All tested parameters were significantly higher in patients than in control subjects. There were no significant differences between patients presenting with or without ST-segment elevation at entry. No correlation was found between time from symptom onset to blood sampling and cytokines levels.
Transcardiac TNF-α, sTMFR-1, and sTNFR-2 Gradients
TNF-α, sTNFR-1, and sTNFR-2 sampled in the aorta (32±8, 1587±441, and 2401±334 pg/mL) did not differ from those in femoral vein (30±10, 1600±440, 2399±302 pg/mL) or in coronary sinus (32±9, 1596±442, 2380±289 pg/mL). In particular, each single measured transcardiac gradient almost always fell within the variability coefficient of the tests used (online figure available at http://www.circulationaha.org).
Eleven patients (6%) died during hospitalization as a result of left (n=4) or right (n=1) ventricle free wall rupture or refractory cardiogenic shock (n=6). Thirty-nine patients (21%) developed new-onset HF. Altogether, 44 patients (24%) satisfied the primary outcome criteria. TNF-α, sTNFR-1, sTNFR-2, IL-6, and IL-1ra were significantly increased in this subgroup (Table 3).
After a median follow-up of 406 days (range, 346 to 696 days), 13 patients (8%) died (12 of cardiovascular causes; 1 of lung cancer), and 24 (14%) had HF, including 18 patients admitted with Killip class >1 at entry who, after complete recovery before discharge, developed HF at follow-up. Cumulatively, 30 patients (18%) satisfied the primary outcome criteria. In these patients, TNF-α, sTNFR-1, sTNFR-2, and IL-6 were increased (Table 3).
Kaplan-Meier estimates of freedom from death and new-onset HF survival at 30 days and 6 and 12 months were 0.79 (95% CI, 0.73 to 0.85), 0.71 (95% CI, 0.64 to 0.78), and 0.69 (95% CI, 0.63 to 0.76), respectively. Overall, 24 patients (13%) died, and 32 (17%) developed new-onset HF. As shown in Table 3, those satisfying the primary outcome were on average older, more often had an anterior necrotic injury or previous MI, and displayed higher white blood cell counts, CRP levels, and peaks of CK, CK-MB, and troponin I, together with a lower LVEF. All tested inflammatory markers were significantly higher in this subgroup compared with those free from death and HF.
In Cox proportional-hazards regression, history of previous MI, Killip class at entry (1 versus >1), LVEF, age, CK-MB at peak, and CRP, together with TNF-α, sTNFR-1, sTNFR-2, IL-6, IL-1ra, and IL-10, were predictors of event-free survival for the composite of death or HF. Use of β-blockers or statins was also associated with better outcome. The hazard ratios (HRs) for these parameters are reported in Table 4. Figure 1 shows differences in primary outcome (Figure 1A) and global survival alone (Figure 1B) when patients were stratified into sTNFR-1 tertiles. In log-rank analysis, there were also significant differences in global survival alone when the first and third tertiles of TNF-α, sTNFR-2, IL-6, and IL-1ra were compared, whereas IL-10 did not reach statistical significance (P=0.057) (data not shown).
By incorporating as putative predictors all significant variables at univariate analysis plus white blood cell count at entry, patients with sTNFR-1 above the median value had an ≈2-fold risk of death and HF (adjusted HR, 1.8; 95% CI, 1.6 to 2.2) compared with those below the median (Table 4); compared with those in the lowest tertile, the risk of death and HF in patients in the highest sTNFR-1 tertile showed an ≈3-fold increase (adjusted HR, 2.9; 95% CI, 1.9 to 3.8). Other independent predictors were LVEF, Killip class, and peak CK-MB, whereas use of statin yielded a borderline significance (P=0.057) (Table 4). sTNFR-1 remained independently associated with primary outcome when the analysis was restricted to STEMI patients or those who underwent primary PCI at entry.
Sensitivity, specificity, positive and negative predictive values, and ROC AUC censored at 30 days and 6 and 12 months for all continuous independent primary outcome predictors at multivariate analysis are reported in Table 5. Overall, when time was not censored, the best cutoff values for sTNFR-1 (1860 pg/mL) had a 73% sensitivity and 72% specificity to predict death and new-onset HF, with a positive and negative predictive value of 71% and 94%, respectively. Cutoff values of 45% and 210 ng/mL maximized sensitivity (66% and 60%) and specificity (76% and 56%) for LVEF and CK-MB, respectively. Positive predictive values were 74% and 52%, whereas negative predictive values were 95% and 88% for the calculated cutoff of LVEF and CK-MB, respectively. ROC AUC for sTNFR-1 (0.82±0.04; 95% CI, 0.73 to 0.90) was higher than that of CK-MB (0.61±0.08; 95% CI, 0.44 to 0.78; P<0.01) but did not differ from that calculated for LVEF (0.82±0.04; 95% CI, 0.73 to 0.92).
To explore the additive prognostic value of combining sTNFR-1 and LVEF, Kaplan-Meier curves were constructed according to the 4 combinations generated by having lower or higher identified overall best cutoffs. As shown in Figure 2, all patients with both LVEF <45% and sTNFR-1>1860 pg/mL satisfied the primary outcome during follow-up, having a worse prognosis than patients with only LVEF below (P<0.02) or sTNFR-1 above (P<0.03) best cutoffs. Multivariate Cox analyses did not show any statistical interaction between the 2 (P=0.3).
Proinflammatory cytokine elevation has been shown to predict short- and long-term incidence of cardiovascular adverse events in several conditions, including HF3–5 and non-STEMI,11 and in large cohort of elderly, well-functioning subjects.6,12 In a nested case-control analysis, TNF-α levels, measured several months after MI, were also shown to predict death and recurrent ischemic events in a low-risk population.13
However, proinflammatory cytokines such as TNF-α are known to be upregulated in the myocardium in response to both MI/necrosis and reperfusion.14,15 Therefore, it remained to be determined whether their elevation during the acute phase of MI simply reflects necrotic injury extent or carries independent prognostic value.
Patients presenting with Killip class 4 at entry or those with history of HF have been intentionally excluded from our investigation to allow us to select a group at low to intermediate risk in whom the clinical relevance of new predictive markers can be better established.
All tested parameters were increased, and in univariate analysis, they all showed to be predictive for the composite end point of death and new-onset HF when in-hospital and follow-up events were combined. However, several baseline and disease-related variables known to have prognostic value such as LVEF7 and age5 can interfere with levels of inflammatory cytokines. Accordingly, results of an unadjusted analysis in this setting should be considered with caution.16 Our multivariate analysis suggests that only levels of sTNFR-1 during acute MI predict the occurrence of death and new-onset HF independently from LV systolic function, Killip class, and peak CK-MB.
To compare independent continuous parameters that emerged in the multivariate approach, the AUCs for LVEF, peak CK-MB, and sTNFR-1 were calculated. LVEF and sTNFR-1 showed the highest combined sensitivity and specificity. Finally, to assess whether the combination of both parameters resulted in a synergistic predictive model, the statistical interaction between these 2 variables was evaluated. Our results could not confirm the existence of an interaction between LVEF and sTNFR-1 in the setting of acute MI. However, the combination of the 2 prognostic elements resulted in a significantly higher capability to predict worse outcome than that provided by each parameter alone. In particular, all patients having LVEF <45% and sTNFR-1>1860 pg/mL satisfied the primary outcome at follow-up.
Interestingly, similar results have been reported for B-type natriuretic peptide and LVEF in a large cohort of MI patients.17 Because our sample size was not powered to assess the statistical interaction between independent predictors, future studies are warranted to specifically address this issue. In addition, in light of our results, it would be of utmost interest to compare the predictive power of B-type natriuretic peptide with that provided by sTNFR-1 in this subset of patients.
In our study, blood sampling was performed early after admission, when β-blockers, aspirin, thienopyridines, nitrates, and ACE inhibitors had just been given and on average had not reached the steady state yet. Therefore, the predictive power of sTNFR-1, when evaluated later during hospitalization or at discharge, warrants further investigations. Because 6 inflammatory biomarkers were investigated simultaneously in an attempt to find the best primary outcome predictor, a limitation of our work is that multiple comparisons have been performed in analyzing our data, which could theoretically increase the type I (α) error. Finally, it remains to be seen whether our findings can be duplicated in a nonwhite population.
The origin of elevated circulating TNF-α and related soluble receptors during acute coronary syndromes and HF is still a matter of debate, because they have been shown to arise both from the inflamed heart18 and from several extracardiac sources, including peripheral skeletal muscles19 and immune system activation.20 No transcardiac or systemic AV difference was found in our study, suggesting a diffuse rather than localized release of the tested cytokines in our patient population.
In keeping with previous results, we found that sTNFR-1 rather than TNF-α antigenic level was a better outcome predictor. It has been shown that at physiological concentrations, sTNFRs may act as a “slow-release reservoir” of bioactive TNF, thus increasing its half-life.21 Because TNF-α induces shedding of its soluble receptors, it is possible that increased sTNFRs simply reflect activation of the cytokine at a local level. In this case, sTNFRs could be sensitive “serum markers” of local TNF-α activation. This hypothesis is supported by the positive correlation between TNF-α antigenic levels and sTNFR-1 in our study (r=0.68, P<0.01).
Alternatively, because several other proinflammatory cytokines are also known to promote sTNFRs shedding,22 sTNFRs could mirror systemic pan-inflammatory status more closely than a single cytokine antigenic level.
In conclusion, our results suggest that, in the setting of acute MI, sTNFR-1 levels carry prognostic information, which is independent from and, at the same time, additive with some well-recognized outcome predictors such as LVEF. Our data extend previous findings on the prognostic role of the TNF-α system in the acute MI setting and reinforce the need to assess whether, in clinical practice, inflammatory biomarkers can be used to tailor optimal medical management of this subset of patients.
The online-only Data Supplement can be found with this article at http://www.circulationaha.org.
Bolognese L, Neskovic AN, Parodi G, Cerisano G, Buonamici P, Santoro GM, Antoniucci D. Left ventricular remodeling after primary coronary angioplasty: patterns of left ventricular dilation and long-term prognostic implications. Circulation. 2002; 106: 2351–2357.
Ferrari R, Bachetti T, Confortini R, Opasich C, Febo O, Corti A, Cassani G, Visioli O. Tumor necrosis factor soluble receptors in patients with various degrees of congestive heart failure. Circulation. 1995; 92: 1479–1486.
Rauchhaus M, Doehner W, Francis DP, Davos C, Kemp M, Liebenthal C, Niebauer J, Hooper J, Volk HD, Coats AJ, Anker SD. Plasma cytokine parameters and mortality in patients with chronic heart failure. Circulation. 2000; 102: 3060–3067.
Deswal A, Petersen NJ, Feldman AM, Young JB, White BG, Mann DL. Cytokines and cytokine receptors in advanced heart failure: an analysis of the cytokine database from the Vesnarinone trial (VEST). Circulation. 2001; 103: 2055–2059.
Vasan RS, Sullivan LM, Roubenoff R, Dinarello CA, Harris T, Benjamin EJ, Sawyer DB, Levy D, Wilson PW, D’Agostino RB. Inflammatory markers and risk of heart failure in elderly subjects without prior myocardial infarction: the Framingham Heart Study. Circulation. 2003; 107: 1486–1491.
Valgimigli M, Agnoletti L, Curello S, Comini L, Francolini G, Mastrorilli F, Merli E, Pirani R, Guardigli G, Grigolato PG, Ferrari R. Serum from patients with acute coronary syndromes displays a proapoptotic effect on human endothelial cells: a possible link to pan-coronary syndromes. Circulation. 2003; 107: 264–270.
Ceconi C, Ferrari R, Bachetti T, Opasich C, Volterrani M, Colombo B, Parrinello G, Corti A. Chromogranin A in heart failure: a novel neurohumoral factor and a predictor for mortality. Eur Heart J. 2002; 23: 967–974.
Guidelines for the diagnosis of heart failure: the Task Force on Heart Failure of the European Society of Cardiology. Eur Heart J. 1995; 16: 741–751.
Mueller C, Buettner HJ, Hodgson JM, Marsch S, Perruchoud AP, Roskamm H, Neumann FJ. Inflammation and long-term mortality after non-ST elevation acute coronary syndrome treated with a very early invasive strategy in 1042 consecutive patients. Circulation. 2002; 105: 1412–1415.
Cesari M, Penninx BW, Newman AB, Kritchevsky SB, Nicklas BJ, Sutton-Tyrrell K, Rubin SM, Ding J, Simonsick EM, Harris TB, Pahor M. Inflammatory markers and onset of cardiovascular events: results from the Health ABC study. Circulation. 2003; 108: 2317–2322.
Ridker PM, Rifai N, Pfeffer M, Sacks F, Lepage S, Braunwald E. Elevation of tumor necrosis factor-α and increased risk of recurrent coronary events after myocardial infarction. Circulation. 2000; 101: 2149–2153.
Valgimigli M, Merli E, Malagutti P, Soukhomovskaia O, Cicchitelli G, Antelli A, Canistro D, Francolini G, Macri G, Mastrorilli F, Paolini M, Ferrari R. Hydroxyl radical generation, levels of tumor necrosis factor-alpha, and progression to heart failure after acute myocardial infarction. J Am Coll Cardiol. 2004; 43: 2000–2008.
Richards AM, Nicholls MG, Espiner EA, Lainchbury JG, Troughton RW, Elliott J, Frampton C, Turner J, Crozier IG, Yandle TG. B-type natriuretic peptides and ejection fraction for prognosis after myocardial infarction. Circulation. 2003; 107: 2786–2792.
Tsutamoto T, Wada A, Matsumoto T, Maeda K, Mabuchi N, Hayashi M, Tsutsui T, Ohnishi M, Sawaki M, Fujii M, Yamamoto T, Horie H, Sugimoto Y, Kinoshita M. Relationship between tumor necrosis factor-alpha production and oxidative stress in the failing hearts of patients with dilated cardiomyopathy. J Am Coll Cardiol. 2001; 37: 2086–2092.
Liuzzo G, Angiolillo DJ, Buffon A, Rizzello V, Colizzi C, Ginnetti F, Biasucci LM, Maseri A. Enhanced response of blood monocytes to in vitro lipopolysaccharide-challenge in patients with recurrent unstable angina. Circulation. 2001; 103: 2236–2241.